gevcdn.cost: Cost function for GEV CDN model fitting

In GEVcdn: GEV Conditional Density Estimation Network

Description

The generalized maximum likelihood (GML) cost function used for GEV CDN model fitting (Martins and Stedinger, 2000). Calculates the negative of the logarithm of the GML, which includes a shifted beta distribution prior for the GEV shape parameter. A normal distribution prior can also be set for the magnitude of the input-hidden layer weights, thus leading to weight penalty regularization.

Usage

gevcdn.cost(weights, x, y, n.hidden, Th, fixed, scale.min, beta.p,
            beta.q, sd.norm)

Arguments

weights	weight vector of length returned by gevcdn.initialize.
x	covariate matrix with number of rows equal to the number of samples and number of columns equal to the number of variables.
y	column matrix of target values with number of rows equal to the number of samples.
n.hidden	number of hidden nodes in the GEV CDN model.
Th	hidden layer transfer function; defaults to gevcdn.logistic.
fixed	vector indicating GEV parameters to be held constant; elements chosen from c("location", "scale", "shape")
scale.min	minimum allowable value for the GEV scale parameter.
beta.p	shape1 parameter for shifted beta distribution prior for GEV shape parameter.
beta.q	shape2 parameter for shifted beta distribution prior for GEV shape parameter.
sd.norm	sd parameter for normal distribution prior for the magnitude of input-hidden layer weights; equivalent to weight penalty regularization.

References

Martins, E.S. and J.R. Stedinger, 2000. Generalized maximum-likelihood generalized extreme-value quantile estimators for hydrologic data. Water Resources Research, 36: 737-744. DOI: 10.1029/1999WR900330

See Also

gevcdn.fit, gevcdn.bag, dgev, optim

GEVcdn documentation built on Dec. 5, 2017, 9:03 a.m.